MPC understanding:
the key point is to find an iteration equation x(i+1)/x(i) propagated by inputs/controls u(i), then optimize the object function J formated by (Xref-X)^2 quadrutic cost functions, then choose the first optimized u(1) value
hyperparameter:
ts --- steps time --- smaller means more precise similar to reference
T --- time horizon --- larger means more early to expect changeing
bicycle cant do side drift that is because of the non-holonomic constraint
diff_car_00: the original diff car model from pranav's robotic class (https://www.youtube.com/watch?v=n8Tda3KFvZc&t=2537s)
diff_car_01: the diff car model without non-holonomic constraints so that vx, vy, omega are independent to each other